From kde-core-devel  Thu Nov 25 14:29:15 2004
From: Waldo Bastian <bastian () kde ! org>
Date: Thu, 25 Nov 2004 14:29:15 +0000
To: kde-core-devel
Subject: Re: RFC: Keeping track of online/offline status & central DNS
Message-Id: <200411251529.19319.bastian () kde ! org>
X-MARC-Message: https://marc.info/?l=kde-core-devel&m=110139289821155
MIME-Version: 1
Content-Type: multipart/mixed; boundary="--nextPart2780550.MfCJv2Z4If"

--nextPart2780550.MfCJv2Z4If
Content-Type: text/plain;
  charset="utf-8"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

> > Proposed DCOP interface
> > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> >
> > enumOnlineStatus { Unknown =3D 0, OffLine =3D 1, Online =3D 2 }
>
> Wouldn't we need additional intermediate states here, something like
> RequestedOffline and RequestedOnline?=20

Would that add value? Are their cases where you would want to treat these=20
differently from the "Offline" state? Then again, probably the same can be=
=20
asked about "Unknown" and "Online".

> It might take a while to get online
> and when going from Online to Offline it would be desirable to wait with
> actually switching off the network until pending network operations are
> finished.

In order to support gracefully going offline we would need to keep track of=
=20
all pending network operations (something slavebase already provides for bt=
w)
=46or the kded module I think it would be enough if applications could info=
rm it=20
about "no network operation active" and "network operation may be active" a=
nd=20
in the latter case establish some "please shutdown" call together with a "o=
k,=20
done" / "please cancel shutdown" reply. A bit like the queryClose stuff tha=
t=20
session management does. If you are going to query the user then there is t=
he=20
"Never mind, the network disconnected itself already" case to keep in mind =
as=20
well :-)

If you want to take "local access" into account as well, it becomes a bit m=
ore=20
tricky because then you also have to solve the problem of figuring out whic=
h=20
connection is affected and which is not.

> > enumOnlineStatus (int?) onlineStatus()
> > // Returns internet status, can be used by applications such as KMail /
> > // KWeather in deciding whether to do certain background activities
> >
> > void setOnlineStatus(enumOnlineStatus (int?) )
> > // Sets internet status, to be used by kppp / kisdn or distribution
> > specific // tools
>
> Is this meant to set the status reflecting the actual network state or to
> request a status change?

It is meant to set the status reflecting the actual network state.

> Or asked in a different way: Would kppp call this function or be called by
> this function?

kppp would call this.

> > setNetworkActivationCallback(DCOPRef, function)
> > // Specify DCOP function that should be called in order to request
> > activation // of network
> > // To be used by kppp / kisdn or distribution specific tools
>
> How would this callback be used?
>
> Might it be desirable to not require DCOP here, so that distribution
> specific tools don't need to be able to use DCOP?

In KDE4 they will be able to use DBUS instead. I don't think it makes sense=
 to=20
develop a new IPC mechanism just to fill the gap till then. Note that they=
=20
can use the dcop command line utility as long as we don't put non-standard=
=20
data types in the DCOP interface. That's a point to keep in mind of course.

Cheers,
Waldo
=2D-=20
bastian@kde.org   |   Free Novell Linux Desktop 9 Evaluation Download
bastian@suse.com  |   http://www.novell.com/products/desktop/eval.html

--nextPart2780550.MfCJv2Z4If
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQBBpew/N4pvrENfboIRAgMtAJ9F9CR3mg9tGCPYHrw2rUCIPDlVkACfYXHy
iZOAZPuyIDSR2FVDsZnNaBc=
=EjZR
-----END PGP SIGNATURE-----

--nextPart2780550.MfCJv2Z4If--